🎨 regenerate receipt v5

Author: ianardee

mindee/documents/receipt/line_item_v5.py

@@ -1,7 +1,6 @@
 from typing import List, Optional, TypeVar
 
 from mindee.documents.base import Document, TypeApiPrediction, clean_out_string
-from mindee.documents.receipt.line_item_v5 import ReceiptV5LineItem
 from mindee.fields.amount import AmountField
 from mindee.fields.classification import ClassificationField
 from mindee.fields.company_registration import CompanyRegistrationField
@@ -10,40 +9,44 @@
 from mindee.fields.tax import Taxes
 from mindee.fields.text import TextField
 
+from .receipt_v5_line_item import ReceiptV5LineItem
+
 
 class ReceiptV5(Document):
-    locale: LocaleField
-    """locale information"""
-    total_amount: AmountField
-    """The total amount paid including taxes, discounts, fees, tips, and gratuity."""
-    date: DateField
-    """The date the purchase was made."""
-    time: TextField
-    """Time of purchase with 24 hours formatting (HH:MM)."""
+    """Receipt v5 prediction results."""
+
     category: ClassificationField
     """The receipt category among predefined classes."""
-    subcategory: ClassificationField
-    """The receipt sub category among predefined classes for transport and food."""
+    date: DateField
+    """The date the purchase was made."""
     document_type: ClassificationField
     """Whether the document is an expense receipt or a credit card receipt."""
-    supplier_name: TextField
-    """The name of the supplier or merchant."""
-    supplier_phone_number: TextField
-    """The Phone number of the supplier or merchant."""
+    line_items: List[ReceiptV5LineItem]
+    """Full extraction of lines, including: description, quantity, unit price and total."""
+    locale: LocaleField
+    """The locale identifier in BCP 47 (RFC 5646) format: ISO language code, '-', ISO country code."""
+    subcategory: ClassificationField
+    """The receipt sub category among predefined classes for transport and food."""
     supplier_address: TextField
-    """The address of the supplier or merchant."""
+    """The address of the supplier or merchant returned as a single string."""
     supplier_company_registrations: List[CompanyRegistrationField]
     """List of supplier company registrations or identifiers."""
+    supplier_name: TextField
+    """The name of the supplier or merchant."""
+    supplier_phone_number: TextField
+    """The Phone number of the supplier or merchant returned as a single string."""
     taxes: Taxes
     """List of tax lines information including: Amount, tax rate, tax base amount and tax code."""
-    total_tax: AmountField
-    """The total amount of taxes."""
-    total_net: AmountField
-    """The total amount excluding taxes."""
+    time: TextField
+    """Time of purchase with 24 hours formatting (HH:MM)."""
     tip: AmountField
     """The total amount of tip and gratuity."""
-    line_items: List[ReceiptV5LineItem]
-    """Full extraction of lines, including: description, quantity, unit price and total."""
+    total_amount: AmountField
+    """The total amount paid including taxes, discounts, fees, tips, and gratuity."""
+    total_net: AmountField
+    """The total amount excluding taxes."""
+    total_tax: AmountField
+    """The total amount of taxes."""
 
     def __init__(
         self,
@@ -52,7 +55,7 @@ def __init__(
         page_n: Optional[int] = None,
     ):
         """
-        Receipt document.
+        Receipt v5 prediction results.
 
         :param api_prediction: Raw prediction from HTTP response
         :param input_source: Input object
@@ -70,67 +73,104 @@ def _build_from_api_prediction(
         self, api_prediction: TypeApiPrediction, page_n: Optional[int] = None
     ) -> None:
         """
-        Build the document from an API response JSON.
+        Build the object from the prediction API JSON.
 
         :param api_prediction: Raw prediction from HTTP response
-        :param page_n: Page number for multi pages pdf input
+        :param page_n: Page number
         """
-        self.locale = LocaleField(api_prediction["locale"], page_id=page_n)
-        self.total_amount = AmountField(api_prediction["total_amount"], page_id=page_n)
-        self.total_net = AmountField(api_prediction["total_net"], page_id=page_n)
-        self.total_tax = AmountField(api_prediction["total_tax"], page_id=page_n)
-        self.tip = AmountField(api_prediction["tip"], page_id=page_n)
-        self.date = DateField(api_prediction["date"], page_id=page_n)
-        self.category = ClassificationField(api_prediction["category"], page_id=page_n)
-        self.subcategory = ClassificationField(
-            api_prediction["subcategory"], page_id=page_n
+        self.category = ClassificationField(
+            api_prediction["category"],
+            page_id=page_n,
+        )
+        self.date = DateField(
+            api_prediction["date"],
+            page_id=page_n,
         )
         self.document_type = ClassificationField(
-            api_prediction["document_type"], page_id=page_n
+            api_prediction["document_type"],
+            page_id=page_n,
         )
-        self.supplier_name = TextField(
-            api_prediction["supplier_name"], value_key="value", page_id=page_n
+        self.line_items = [
+            ReceiptV5LineItem(prediction, page_id=page_n)
+            for prediction in api_prediction["line_items"]
+        ]
+        self.locale = LocaleField(
+            api_prediction["locale"],
+            page_id=page_n,
         )
-        self.supplier_phone_number = TextField(
-            api_prediction["supplier_phone_number"], value_key="value", page_id=page_n
+        self.subcategory = ClassificationField(
+            api_prediction["subcategory"],
+            page_id=page_n,
         )
         self.supplier_address = TextField(
-            api_prediction["supplier_address"], value_key="value", page_id=page_n
+            api_prediction["supplier_address"],
+            page_id=page_n,
         )
         self.supplier_company_registrations = [
-            CompanyRegistrationField(field_dict, page_id=page_n)
-            for field_dict in api_prediction["supplier_company_registrations"]
+            CompanyRegistrationField(prediction, page_id=page_n)
+            for prediction in api_prediction["supplier_company_registrations"]
         ]
-        self.time = TextField(api_prediction["time"], value_key="value", page_id=page_n)
+        self.supplier_name = TextField(
+            api_prediction["supplier_name"],
+            page_id=page_n,
+        )
+        self.supplier_phone_number = TextField(
+            api_prediction["supplier_phone_number"],
+            page_id=page_n,
+        )
         self.taxes = Taxes(api_prediction["taxes"], page_id=page_n)
-        self.line_items = [
-            ReceiptV5LineItem(prediction=line_item, page_n=page_n)
-            for line_item in api_prediction["line_items"]
-        ]
+        self.time = TextField(
+            api_prediction["time"],
+            page_id=page_n,
+        )
+        self.tip = AmountField(
+            api_prediction["tip"],
+            page_id=page_n,
+        )
+        self.total_amount = AmountField(
+            api_prediction["total_amount"],
+            page_id=page_n,
+        )
+        self.total_net = AmountField(
+            api_prediction["total_net"],
+            page_id=page_n,
+        )
+        self.total_tax = AmountField(
+            api_prediction["total_tax"],
+            page_id=page_n,
+        )
 
     @staticmethod
-    def _line_items_separator(char: str):
+    def _line_items_separator(char: str) -> str:
         out_str = "  "
         out_str += f"+{char * 38}"
         out_str += f"+{char * 10}"
         out_str += f"+{char * 14}"
         out_str += f"+{char * 12}"
         return out_str + "+"
 
+    def _line_items_to_str(self) -> str:
+        if not self.line_items:
+            return ""
+
+        lines = f"\n{self._line_items_separator('-')}\n  ".join(
+            [item.to_table_line() for item in self.line_items]
+        )
+        out_str = ""
+        out_str += f"\n{self._line_items_separator('-')}\n "
+        out_str += " | Description                         "
+        out_str += " | Quantity"
+        out_str += " | Total Amount"
+        out_str += " | Unit Price"
+        out_str += f" |\n{self._line_items_separator('=')}"
+        out_str += f"\n  {lines}"
+        out_str += f"\n{self._line_items_separator('-')}"
+        return out_str
+
     def __str__(self) -> str:
         supplier_company_registrations = "; ".join(
             [str(n.value) for n in self.supplier_company_registrations]
         )
-        line_items = "\n"
-        if self.line_items:
-            line_items = (
-                f"\n{self._line_items_separator('-')}"
-                "\n  | Description                          | Quantity | Total Amount | Unit Price |"
-                f"\n{self._line_items_separator('=')}"
-            )
-            for item in self.line_items:
-                line_items += f"\n  {item}\n{self._line_items_separator('-')}"
-
         return clean_out_string(
             "Receipt V5 Prediction\n"
             "=====================\n"
@@ -150,11 +190,8 @@ def __str__(self) -> str:
             f":Supplier Company Registrations: {supplier_company_registrations}\n"
             f":Supplier Address: {self.supplier_address}\n"
             f":Supplier Phone Number: {self.supplier_phone_number}\n"
-            f":Line Items: {line_items}\n"
+            f":Line Items: {self._line_items_to_str()}\n"
         )
 
-    def _checklist(self) -> None:
-        pass
-
 
 TypeReceiptV5 = TypeVar("TypeReceiptV5", bound=ReceiptV5)

mindee/documents/receipt/receipt_v5.py

@@ -0,0 +1,77 @@
+from typing import Dict, Optional
+
+from mindee.fields.base import (
+    FieldConfidenceMixin,
+    FieldPositionMixin,
+    TypePrediction,
+    float_to_string,
+    to_opt_float,
+)
+
+
+class ReceiptV5LineItem(FieldPositionMixin, FieldConfidenceMixin):
+    """Full extraction of lines, including: description, quantity, unit price and total."""
+
+    description: Optional[str]
+    """The item description."""
+    quantity: Optional[float]
+    """The item quantity."""
+    total_amount: Optional[float]
+    """The item total amount."""
+    unit_price: Optional[float]
+    """The item unit price."""
+    confidence: float = 0.0
+    """Confidence score"""
+    page_n: int
+    """The document page on which the information was found."""
+
+    def __init__(
+        self,
+        prediction: TypePrediction,
+        page_id: Optional[int] = None,
+    ):
+        self._set_confidence(prediction)
+        self._set_position(prediction)
+
+        if page_id is None:
+            try:
+                self.page_n = prediction["page_id"]
+            except KeyError:
+                pass
+        else:
+            self.page_n = page_id
+
+        self.description = prediction["description"]
+        self.quantity = to_opt_float(prediction, "quantity")
+        self.total_amount = to_opt_float(prediction, "total_amount")
+        self.unit_price = to_opt_float(prediction, "unit_price")
+
+    def _printable_values(self) -> Dict[str, str]:
+        """Return values for printing."""
+        return {
+            "description": self.description if self.description is not None else "",
+            "quantity": float_to_string(self.quantity),
+            "total_amount": float_to_string(self.total_amount),
+            "unit_price": float_to_string(self.unit_price),
+        }
+
+    def to_table_line(self) -> str:
+        """Output in a format suitable for inclusion in an rST table."""
+        printable = self._printable_values()
+        return (
+            "|"
+            f" {printable['description']:<36} |"
+            f" {printable['quantity']:<8} |"
+            f" {printable['total_amount']:<12} |"
+            f" {printable['unit_price']:<10} |"
+        )
+
+    def __str__(self) -> str:
+        """Default string representation."""
+        printable = self._printable_values()
+        return (
+            f"Description: {printable['description']}, "
+            f"Quantity: {printable['quantity']}, "
+            f"Total Amount: {printable['total_amount']}, "
+            f"Unit Price: {printable['unit_price']}, "
+        ).strip()